vapourminer (OP)
Legendary
 Offline
Activity: 4326
Merit: 3536
what is this "brake pedal" you speak of?
|
 |
August 18, 2012, 01:00:19 PM
Last edit: August 18, 2012, 07:51:40 PM by vapourminer |
|
did a search here and google and perhaps my search foo sucks but I cant find anything on this. what exactly are the differences?
my single arrived. packing slip indicates "Rev 1".
on top, the unit has the gold and silver HSF with the heat pipes. the bottom has a small fan also. so isnt this a revision 3? revision 1 (the originals) had a solid black HSF with no heat pipes on top didnt they?
auto tune settled on an 816 MH/s bitstream. cgminer says 800 average. @ 73 F ambient, this is withing spec, but Ill pull it apart and redo the heat sink compound with MX 2 later today and hopefully pull better performance.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Once a transaction has 6 confirmations, it is extremely unlikely that an attacker without at least 50% of the network's computation power would be able to reverse it.
|
|
|
Advertised sites are not endorsed by the Bitcoin Forum. They may be unsafe, untrustworthy, or illegal in your jurisdiction.
|
|
|
KIDC
Member
Offline
Activity: 80
Merit: 10
|
|
August 19, 2012, 12:57:16 AM |
|
I was confused by this also. My paperwork all states revision 1 too. I have 6 Singles, one with three fans and the other five with two. To my knowledge all the currently shipping singles have a small 40mm or so fan inside a custom heatsink attached to the bottom of the PCB. It's not easy to see since it's inside the enclosure, just above the bottom grill. All units have a top 92mm fan that pulls air through the unit from the side vents and pushes it out the top. A few of them have a third fan, also 92mm, on the bottom as well, which also pulls air out of the unit from the side vents. This is because a few units need this extra airflow to remain stable.
I think the "Rev 1." listed on the sheet that's included by BFL in the shipments is an oversight. They may be referring to BitForce version 1.0 perhaps, or may simply have not realized to change it. To my knowledge there are three different revisions. This was mentioned months ago by a BFL rep. in a thread somewhere I believe, but it's not easy to find. If my memory serves:
Rev 1: internal BFL only testing device (also briefly tested by Inaba aka BFL_Josh briefly also I think). Was not stable with the black smaller heatsink and lone 92mm fan it used. Had smaller vents on the side to I think. This unit had serious throttling issues I believe.
Rev 2: first product shipped to customers (early adopters right at top of wait list). Still had throttling issues. Had the same heatsink array but had two 92mm fans, one of which was hastily added to the bottom.
Rev 3: all the current shipping singles. I believe it's only Rev 3. that has the large copper/aluminum heatsink array that you have and I have on all my units. This started shipping in March I think.
On all my Singles I have removed the front/side metal panel (just has a few screws, easily done) and flipped the fans so that they push air down onto the heatsinks instead. This seems to reduce temperatures by a few degrees celsius.
Be careful with the actual thermal compound. It's actually two FPGAs which may or may not be perfectly level. I don't know that redoing the thermal paste will help the situation. I'm not super competent with thermal compound and will be trading in these units for ASICs before long anyway so I did not bother with any regreasing attempts. Be careful with auto-tune too. It wasn't actually accurate for me for 24/7 operation. Also keep in mind that cgminer will always show an average of a little less than the actual firmware that's on the device, as you're already experiencing with 800/816, so don't be alarmed by that. Keep an eye on the front LED blinking as well as the HW count in cgminer, as blinking LED usually indicates throttling (but only blinks while it's doing it), and the recent few versions of cgminer are supposed to count throttles on BFL hardware under the HW count. I haven't actually tested this myself.
Hope that helps.
|
"He who controls the past commands the future. He who commands the future, conquers the past."
|
|
|
|
Askit2
|
|
August 19, 2012, 03:06:30 AM |
|
Very nearly correct.
Rev 1. First units shipped black or whatever heatsink. Was publicly sent out to people who paid for them.
Rev 2. Higher temp FPGA shipped later then the first ones. Upgraded part sent out about when the 2-3 week delay happened on shipping posts.
Rev 3. 2-3 fans total with heatpipes and supposedly higher temp FPGAs. Slight delay but not a large one.
|
|
|
|
vapourminer (OP)
Legendary
Offline
Activity: 4326
Merit: 3536
what is this "brake pedal" you speak of?
|
|
August 19, 2012, 06:25:03 PM |
|
KIDC, Askit2, thanks for the info.
I wonder if they just keep sending a copy of the same original packing slip as mine has no date.
easyminer mentioned a version number too. could of sworn it was "1" but I cant remember if it said "version" or "revision." Ill check it out next time I run it.
I did flip the fan to blowing down but left the front/side cover on (cats here). temp went up by 1.5 C, but again I left the case together. so its back to blowing up. once I move it to the basement and its protected from the animals Ill run it with no front/sides on it and try the fan both ways.
still gonna change the paste to MX-2 but when I was about to pull the HS I didnt think I had enough for two chips left. Once I get more Ill give it a shot. I like artic cooling paste, it hasnt failed to give better temps on anything Ive put it on yet.
I moved the single to a slightly different location as I think it was getting a bit of warm exhaust air from rig that was driving it. I flashed the 832 bitstream back to it and after 18 hrs no throttles or HW errors so Ill leave it for now. cgminer says 820 MH/s now.
that sucker sure is sensitive to ambient temps but Im not surprised as the case airflow on that thing sucks beyond belief. it needs vents in the upper front next to the HS the heat pipes go through.
|
|
|
|
|
|
abeaulieu
|
|
August 20, 2012, 02:32:28 PM |
|
did a search here and google and perhaps my search foo sucks but I cant find anything on this. what exactly are the differences?
my single arrived. packing slip indicates "Rev 1".
on top, the unit has the gold and silver HSF with the heat pipes. the bottom has a small fan also. so isnt this a revision 3? revision 1 (the originals) had a solid black HSF with no heat pipes on top didnt they?
auto tune settled on an 816 MH/s bitstream. cgminer says 800 average. @ 73 F ambient, this is withing spec, but Ill pull it apart and redo the heat sink compound with MX 2 later today and hopefully pull better performance.
I believe all packing clips read "Rev 1" I just got a Revision 3 single like you did (custom heat sink with heat pipes, bottom mounted small fan + heat sink) and the slip was marked "Rev 1" as well. Just a typo that they haven't updated the document. |
|
|
|
|
fortyniner
Newbie
Offline
Activity: 27
Merit: 0
|
|
August 28, 2012, 11:43:42 PM |
|
Got my first single Friday (shipped 8/22) and it's marked "BitFORCE - 1.0 Rev: E" with big fan & copper tubed heatsink on top, and smaller fan & heatsink on bottom. Oddly, it shipped with 872 firmware but would throttle error every minute or so. Took half the case off, propped it up a little higher and it's fairly stable now with 832mh firmware.
Watching the temps in Cgminer, I've not seen it go over 48.2C. Anything higher seems to trigger throttling on mine (which stops the temp reads for 15 seconds). Temperature appears to cycle up & down half a degree or so. Do you suppose Cgminer is cycling back & forth reading each of the two chips?
I see there's a lot of variation in these singles - some hash high at low temps and some hash low at high temps with others in between. It would appear the overheat/throttle trigger temp is variable and built-in to the chip. In other words, luck of the draw.
|
|
|
|
|
fortyniner
Newbie
Offline
Activity: 27
Merit: 0
|
|
August 31, 2012, 04:35:25 AM |
|
I took it totally out of the case yesterday and today noticed one of the copper fins on the upper heatsink was hung up on the corner screw. That prevented the heatsink from forming a tight, flush union with the 2 chips (may have happened during shipment). I bent the fin out of the way (yeah I should take a pic) and now the heatsink can free-float on the chips, as designed. It runs a degree or 2 cooler now.
|
|
|
|
|
|
HDSolar
|
|
August 31, 2012, 04:56:29 AM |
|
I took it totally out of the case yesterday and today noticed one of the copper fins on the upper heatsink was hung up on the corner screw. That prevented the heatsink from forming a tight, flush union with the 2 chips (may have happened during shipment). I bent the fin out of the way (yeah I should take a pic) and now the heatsink can free-float on the chips, as designed. It runs a degree or 2 cooler now.
Yes, lots of user tweeks needed to make them run well. I hope the SC's are a bit more quality in their production.  |
|
|
|
|
